Supplement of Atmos. Chem. Phys., 19, 14477–14492, 2019 https://doi.org/10.5194/acp-19-14477-2019-supplement © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License.

Supplement of Quantifying the impact of synoptic circulation patterns on ozone variability in northern China from April to October 2013–2017

Jingda Liu et al.

Correspondence to: Lili Wang ([email protected])

The copyright of individual parts of the supplement might differ from the CC BY 4.0 License. 1 Text S1 Lamb-Jenkinson circulation typing approach

2 In synoptic climatology, along with subjective or manual approaches, objective or automated 3 approaches are important and widely used synoptic weather typing procedures to identify recurring 4 map patterns or variable groups that typify significant modes of circulation and to classify each case 5 into one of these modes (Yarnal, 1993; Huth et al., 2008). . There are many objective methods, such as 6 correlation-based map-pattern technique, sums-of-squares method, rotated principal component 7 analysis, hierarchical clustering (average linkage or Ward’s clustering), and K-means clustering. As 8 suggested by Huth (1996)and Yarnal (1993), , all the methods proved to be capable of yielding 9 meaningful classifications and none of them can be thought of as the best in all aspects. Which method 10 to use will depend mainly on the aim of the classification. Notably, the final number of synoptic types 11 using these algorithms is associated with a given period and region, statistical algorithms, prior 12 knowledge of the synoptic climatology of the region, and experimentation with various statistical 13 procedures; finally, a subjective decision as to how many clusters are appropriate for the study period is 14 made by the investigator. Thus, the results of a synoptic-type analysis are quite subjective. 15 The noted British climatologist Lamb developed a synoptic-scale, daily weather-map classification 16 for use over the British Isles, and seven basic types were identified manually (Lamb, 1972). Based on 17 Lamb's study, Jenkinson (1977) improved the subjective approach to an objective approach, called the 18 Lamb-Jenkinson method (Jones et al., 1993; Trigo and DaCamara, 2000). According to the sea level 19 pressure (SLP) of these 16 grids, a set of indices related to the direction and vorticity of geostrophic 20 flow are calculated to determine the weather type. The indices used are the following: southerly flow 21 component of the geostrophic surface wind (SF), westerly flow component of the geostrophic surface 22 wind (WF), resultant flow (FF), southerly shear vorticity (ZS), westerly shear vorticity (ZW) and total 23 shear vorticity (Z). These indices were computed using SLP values obtained for the retained number of 24 grid points, and are both for the flow units as for the geostrophic vorticity expressed in hPa. As shown 25 in 1a, the research area is placed in the central position, which refers to the area of connecting with P4, 26 P8, P12, P13, P9 and P5. The SLP of 16 grids can be used to characterize the distance of between the 27 study region and high-/low- pressure system, which result in the method is available to classify the 28 weather pattern for each day and successfully applied in many areas (Lamb, 1972; Jenkinson, 1977; 29 Trigo and DaCamara, 2000; Demuzere et al., 2009; Santurtún et al., 2015; Pope et al., 2016; Liao et al., 30 2017). The following presents the calculation methods for each index: 31 SF=1.035×[0.25×(P5+2×P9+P13)-0.25×(P4+2×P8+P12)] 32 WF=[0.5×(P12+P13)-0.5×(P4+P5)] 33 ZS=0.85×[0.25(P6+2×P10+P14)-0.25×(P5+2×P9+P13)-0.25×(P4+2×P8+P12)+0.25×(P3+2×P7+P11)] 34 ZW=1.12×[0.5×(P15+P16)-0.5×(P8+P9)]-0.91×[0.5×(P8+P9)-0.5×(P1+P2)] 35 F=(SF2+WF2)1/2 36 Z=ZS+ZW 37 P represents the SLP at the grid point. The positions of 16 grid points are shown in Fig. 1a; for example, 38 P1 is the SLP at the 1st grid point. 39 The weather types are defined by comparing values of FF and Z:

40 (1) Direction of flow (in degrees) is given by tan−1(WF/SF), 180° being added if WF is positive. The

41 appropriate wind direction is computed using an eight-point compass, allowing 45° per sector. 42 (2) If |Z| < FF, flow is essentially straight and considered to be of a pure directional type (eight 43 different possibilities according to the compass directions). 44 (3) If |Z| > 2FF, the pattern is considered to be of a pure cyclonic type if Z > 0 or of a pure anticyclonic 45 type if Z < 0. 46 (4) If FF < |Z| < 2FF, flow is considered to be of a hybrid type and is therefore characterized by both 47 direction and circulation (16 different types). 48 Thus, compared with other objective synoptic classification approaches, the advantage of 49 Lamb-Jenkinson method is that the number of synoptic types and the weather type that is present each 50 day in the specific region is robust and fixed. In addition, the method clearly gives the typical pressure 51 fields (anticyclone, cyclone, directional types and hybrid types), which can well reflect the wind fields 52 over the study region. Particularly, directional types can represent the prevailing wind direction in this 53 area under the control of the specific weather pattern. Many studies have shown that the high/low 54 concentrations of ozone are always associated with the southerly/northerly winds in North China (Han 55 et al., 2019; Li et al., 2019). Consequently, the Lamb-Jenkinson weather type scheme is a better

56 method for exploring the O3 pollution in North China.

57 Text S2 Segmented synoptic-regression approach

58 In an attempt to objectively define and weight the meteorological variables, Eder et al. (1994) indicated

59 that most influence the O3 concentrations within each weather type, and they developed stepwise linear 60 regression models of the following form: 푛 61 푂3 = 훽0 + ∑푖=1 훽푖푥푖 (1)

62 where O3 represents the predicted MDA8 O3 for each weather category, 훽0 is a constant, 훽푖 represents 63 the coefficients (determined by the least squares method) of the independent meteorological variables

64 푥푖, and n is the number of independent meteorological variables in the equation. A stepwise regression 65 procedure was utilized because it sequentially identifies the ‘best subset’ of the independent 66 meteorological variables. 67 This paper used multiple linear regression with a stepwise method for variable selection to reconstruct

68 the time series for O3 with F probability <0.05 to enter and F probability <0.10 to exit. After excluding 69 the missing data and disordering the time sequences, 80% of these days were used to build the potential 70 forecast equations and the remaining 20% were used to validate the accuracy of the equations.

71 Tables

72 Table S1. Locations, period of available data, average MDA8 O3 concentrations, and exceedance ratios of 58 73 cities. Abb, Lon, Lat and Alt are short for abbreviation, longitude, latitude and altitude, respectively.

Province City name Abb Lon Lat Alt Period O3 (μg Exceedance m-3) Ratio (%) Hebei Chengde CD 117.91 40.95 379.20 2013-2017 118 18.93 Hebei Zhangjiakou ZJK 114.87 40.81 860.20 2013-2017 114 15.84 Shanxi Datong DT 113.31 40.10 1051.17 2015-2017 112 9.38 Beijing Beijing BJ 116.38 40.04 58.25 2013-2017 127 29.43 Hebei Qinhuangdao QHD 119.60 39.93 2.80 2013-2017 102 9.56 Hebei Tangshan TS 118.18 39.63 6.83 2013-2017 124 25.12 Hebei Langfang LF 116.71 39.54 19.50 2013-2017 120 22.87 Shanxi Shuozhou SZ 112.43 39.34 1094.60 2015-2017 133 23.13 Tianjin Tianjin TJ 117.35 39.09 6.27 2013-2017 113 17.24 Hebei Baoding BD 115.47 38.87 15.00 2013-2017 125 25.68 Shanxi Xinzhou XZ 112.72 38.45 778.00 2015-2017 110 13.13 Hebei Cangzhou CZ 116.86 38.30 14.33 2013-2017 127 25.59 Hebei Shijiazhuang SJZ 114.48 38.03 153.75 2013-2017 116 21.18 Shanxi Yangquan YQ 113.56 37.86 767.83 2015-2017 119 20.63 Shanxi Taiyuan TY 112.52 37.86 800.89 2013-2017 116 18.67 Hebei Hengshui HS 115.68 37.74 23.00 2015-2017 138 34.86 Shanxi Jinzhong JZ 112.73 37.70 796.00 2015-2017 110 16.25 Shanxi Lvliang LL 111.13 37.51 943.00 2015-2017 88 3.75 Shandong Yantai YT 121.37 37.50 9.17 2015-2017 118 12.50 Shandong Weihai WH 122.09 37.49 26.00 2015-2017 115 9.53 Shandong Dezhou DZ 116.30 37.46 22.67 2015-2017 146 40.78 Shandong Dongying DY 118.64 37.43 3.25 2015-2017 146 35.94 Shandong Binzhou BZ 118.00 37.38 16.67 2015-2017 116 17.19 Hebei Xingtai XT 114.49 37.06 71.75 2013-2017 115 19.96 Shandong Zibo ZB 118.01 36.73 81.67 2015-2017 139 32.03 Shandong Weifang WF 119.14 36.71 53.80 2015-2017 142 34.38 Shandong Jinan JN 116.98 36.65 55.25 2015-2017 130 45.31 Hebei Handan HD 114.51 36.60 64.50 2013-2017 116 31.25 Shandong Liaocheng LC 115.99 36.45 44.67 2015-2017 133 29.06 Shandong Laiwu LW 117.69 36.21 207.67 2015-2017 123 21.56 Shandong Tai’an TA 117.11 36.19 170.00 2015-2017 126 22.81 Shanxi Changzhi CZH 113.11 36.19 924.40 2015-2017 141 36.72 Shandong Qingdao QD 120.39 36.15 23.78 2015-2017 105 12.66 Henan Anyang AY 114.37 36.09 84.20 2015-2017 119 22.03 Shanxi Linfen LFE 111.51 36.08 466.33 2015-2017 115 19.84 Henan Hebi HB 114.25 35.79 115.67 2015-2017 123 19.38 Henan Puyang PY 115.04 35.77 56.00 2015-2017 124 22.19 Shanxi Jincheng JC 112.85 35.50 743.17 2015-2017 103 15.94 Shandong Jining JN 116.59 35.42 34.00 2015-2017 135 31.09 Shandong Rizhao RZ 119.51 35.41 31.00 2015-2017 122 16.09 Henan Xinxiang XX 113.88 35.29 75.50 2015-2017 125 27.81 Shandong Heze HZ 115.45 35.25 47.00 2015-2017 131 25.00 Henan Jiaozuo JZ 113.23 35.22 93.00 2015-2017 122 23.91 Shanxi Yuncheng YC 111.02 35.05 380.20 2015-2017 117 19.06 Shandong Linyi LY 118.33 35.05 69.50 2015-2017 135 29.69 Shandong Zaozhuang ZZH 117.52 34.82 71.40 2015-2017 133 28.13 Henan Kaifeng KF 114.34 34.79 79.25 2015-2017 114 15.47 Henan Sanmenxia SMX 111.17 34.79 353.25 2015-2017 121 19.06 Henan Zhengzhou ZZ 113.65 34.78 111.44 2013-2017 101 17.34 Henan Luoyang LY 112.42 34.66 133.29 2015-2017 127 27.66 Henan Shangqiu SQ 115.66 34.42 33.25 2015-2017 124 18.44 Henan Xuchang XC 113.81 34.01 71.00 2015-2017 124 20.00 Henan Pingdingshan PDS 113.28 33.73 75.75 2015-2017 129 25.78 Henan Zhoukou ZK 114.66 33.61 45.75 2015-2017 116 13.28 Henan Luohe LH 114.03 33.57 65.75 2015-2017 127 22.81 Henan Nanyang NY 112.53 32.99 124.00 2015-2017 128 22.97 Henan Zhumadian ZMD 114.01 32.99 78.00 2015-2017 124 21.09 Henan Xinyang XY 114.06 32.12 89.50 2015-2017 110 10.31 74 Table S2 Days of occurrence and the proportion of the most-polluted synoptic categories (LP and C) in May 75 and the second half of May in each weather category in different years.

N-E-S S-W-N Proportion Year LP C A directions directions (%) 2013 5 14 4 4 4 25.8 2014 7 16 3 4 1 22.6 2015 7 10 4 4 6 25.8 May 2016 8 12 2 2 7 12.9 2017 5 13 3 8 2 35.5 mean 6.4 13 3.2 4.4 4 24.5 2013 2 8 2 2 2 25.0 2014 1 11 2 2 0 25.0 The second 2015 3 7 3 0 3 18.8 half of May 2016 5 6 2 0 3 12.5 2017 2 5 1 7 1 50.0 mean 2.6 7.4 2 3.7 2.25 35.6 76 77 Table S3 Equations of the multiple stepwise regressions for each weather type in 14 cities, where O3 indicates

78 MDA8 O3.

Categories Cities Regression equation CD O3=4.77Tmax+9.97V+1.77ws-0.78RH_lag-1.16Tmax_lag+6.77V_lag+62.81 ZJK O3=2.1Tmax-5.26U-0.75RH_lag+0.79rain_lag+4.7V_lag+94.78 BJ O3=6.71Tmax+15.39V+12.22ws-0.66RH_lag-1.69Tmax_lag-9.49 QHD O3=2.2Tmax+11.25V+4.63 TS O3=-0.76RH+4.93Tmax+4.65U+17.05V+34.84 LF O3=-0.48RH+5.2Tmax+1.66V+1.87 N-E-S TJ O3=4.69Tmax+3.96V-0.4RH_lag+5.53 direction BD O3=4.76Tmax+7.53V-0.49RH_lag-8.15U_lag+7.73V_lag+24.84 CZ O3=-0.57RH+4.73Tmax+8.69V-0.6wd_lag+41.47 SJZ O3=4.9Tmax-6U-0.54RH_lag+28.42 TY O3=-0.41RH+2.53Tmax-0.92pre-3.62U+973.36 XT O3=6.22Tmax-5.14U-0.81RH_lag-1.82Tmax_lag+45.59 HD O3=-0.48RH+4.91Tmax+8.48 ZZ O3=3.47Tmax+1.89Tmax_lag+0.88rain_lag-55.75 CD O3=5.31Tmax-0.64RH_lag+8.89V_lag-1.53pre_lag+1579.54 ZJK O3=5.31Tmax-0.64RH_lag+8.89V_lag-1.53pre_lag+1579.54 BJ O3=7.86Tmax-7.72U+17.80V-79.97 QHD O3=-0.55RH+3.87Tmax+2.85V+48.73 TS O3=6.24Tmax+4.60U+10.64V-0.48RH_lag-0.06wd_lag+6.71 LF O3=6.7Tmax-1.66rain-0.46RH_lag+7.7V_lag-24.83 S-W-N TJ O3=5.59Tmax+6.67V+4.67V_lag-43.81 direction BD O3=7.26Tmax-13.24ws-0.85RH_lag+18.28 CZ O3=5.16Tmax+4.86V-0.71RH_lag-0.7wd_lag+45.20 SJZ O3=6.25Tmax-11.2U-43.55 TY O3=-0.79RH+2.13Tmax-0.7wd-1.9pre-1.23U+212.60 XT O3=4.87Tmax-1.23rain+8.67ws-28.72 HD O3=4.61Tmax-1.54rain+5.15ws_lag-5.27 ZZ O3=4.58Tmax-2.64Tmax_lag+43.43 CD O3=4.39Tmax+15.62V-0.67RH_lag+56.36 ZJK O3=3.6Tmax+9.46V-0.8wd-0.33RH_lag+81.53 BJ O3=7.14Tmax+18.59V-0.49RH_lag-32.17 QHD O3=3.33Tmax+22.46V+14.4ws+11.75 TS O3=-0.88RH+7.83Tmax+18.48V+11.00ws_lag+1.45pre_lag-1511.94 LF O3=8.72Tmax+1.36V+3.27pre-0.74RH_lag-7.47U_lag-3365.76 TJ O3=-0.5RH+6.2Tmax+6.45V-23.85 LP BD O3=-0.93RH+9.33Tmax+3.86pre-3951.69 CZ O3=-1.17RH+3.15Tmax-0.53rain-7.58ws_lag+142.22 SJZ O3=8.52Tmax-12.5U+2.65pre_lag-2785.79 TY O3=7.17Tmax-0.86RH_lag-14.71U_lag-1.68ws_lag-29.64 XT O3=7.7Tmax+2.22pre-2319.11 HD O3=5.88Tmax+2.67pre-2727.43 ZZ O3=3.33Tmax-22.58ws+44.13 CD O3=6.42Tmax+8.27V-0.97RH_lag+0.34rain_lag-11.62U_lag+1.75pre_lag-1746.67 C ZJK O3=4.16Tmax-8.24U+13.91 BJ O3=7.80Tmax+17.77V-0.49RH_lag-47.88 QHD O3=-0.99RH+3.2Tmax-2.33pre+246.27 TS O3=-1.15RH+8.10Tmax-2.13Tmax_lag+11.19V_lag+44.01 LF O3=8.47Tmax+9.52V-0.55RH_lag-75.33 TJ O3=-0.72RH+6.35Tmax+4.93V-13.34 BD O3=-0.78RH+6.35Tmax+7.89V+19.22 CZ O3=-1.59RH+4.39Tmax-5.45U+115.54 SJZ O3=-0.65RH+5.62Tmax-15.31U+8.26 TY O3=6.32Tmax-7.53U-69.28 XT O3=-0.51RH+6.26Tmax+0.12wd+6.37ws_lag-49.57 HD O3=-0.96RH+4.15Tmax+72.43 ZZ O3=6.29Tmax+2pre-289.95 CD O3=-0.79RH+3.19Tmax+9.43V+8.73ws-1.73pre+5.83V_lag+1835.32 ZJK O3=-0.35RH+3.29Tmax+4.23V+6.17ws+4.26V_lag+4ws_lag+39.25 BJ O3=5.59Tmax-1.48pre-0.38RH_lag+1507.66 QHD O3=3.5Tmax+13.9V+12.17V_lag+7.48ws_lag+14.79 TS O3=5.83Tmax-0.66RH_lag+7.17V_lag+16.91 LF O3=6.45Tmax-44.57 TJ O3=6.47Tmax+6.99ws-0.43729RH_lag+3.58V_lag+1.37pre_lag-1433.13 A BD O3=6.69Tmax-0.48RH_lag-13.34 CZ O3=7.64Tmax-0.63RH_lag-3.399Tmax_lag+7.11V_lag+51.58 SJZ O3=-0.43RH+6.19Tmax-16.88 TY O3=4.98Tmax+6.74ws_lag-38.97 XT O3=-0.39RH+5.77Tmax+3.228ws_lag-16.28 HD O3=-0.56RH+6.1Tmax-8.64 ZZ O3=4.57Tmax-0.6wd-1.394pre_lag+1399.78 79 Table S4. Three statistical measures (R2, rmse, and CV) for the building and validation datasets for 5 weather 80 categories and composite model for all 14 cities. Com: composite model (integration of each category’s 81 modeling results). R2 (%): variance in the individual model’s coefficients of determination; rmse (μg m-3):

82 root mean square error; and CV (%): coefficient of variation defined as rmse/mean MDA8 O3.

Building Validation Cities Parameter com HP UP LP C A com HP UP LP C A R2 0.64 0.56 0.60 0.51 0.61 0.67 0.58 0.54 0.73 0.45 0.60 0.51 CD RMSE 29.04 27.77 33.18 33.04 24.70 28.24 32.20 28.19 35.88 33.85 31.45 31.40 CV 0.24 0.28 0.25 0.24 0.19 0.28 0.28 0.29 0.27 0.28 0.25 0.32 R2 0.53 0.36 0.56 0.45 0.31 0.62 0.61 0.47 0.83 0.34 0.89 0.68 ZJK RMSE 29.48 28.25 32.38 31.54 32.34 24.47 34.01 26.78 40.77 44.94 28.03 28.70 CV 0.26 0.29 0.26 0.25 0.26 0.24 0.30 0.30 0.31 0.39 0.22 0.29 R2 0.64 0.57 0.65 0.54 0.59 0.64 0.69 0.67 0.73 0.67 0.47 0.65 BJ RMSE 36.28 35.07 37.62 42.32 36.94 33.28 37.44 36.49 33.42 42.00 39.74 38.71 CV 0.29 0.35 0.27 0.29 0.25 0.31 0.29 0.35 0.23 0.28 0.28 0.36 R2 0.36 0.24 0.33 0.33 0.24 0.43 0.40 0.37 0.54 0.36 0.23 0.46 QHD RMSE 34.25 31.64 34.88 39.31 40.03 28.51 36.29 29.71 37.94 44.32 38.58 33.75 CV 0.34 0.36 0.32 0.33 0.35 0.32 0.36 0.38 0.31 0.42 0.36 0.36 R2 0.62 0.57 0.52 0.68 0.53 0.57 0.57 0.70 0.41 0.53 0.59 0.48 TS RMSE 34.63 33.36 36.29 36.45 37.76 32.40 38.05 35.46 37.23 41.12 32.36 45.38 CV 0.28 0.34 0.26 0.26 0.27 0.30 0.30 0.33 0.27 0.28 0.25 0.36 R2 0.63 0.61 0.60 0.60 0.60 0.65 0.63 0.89 0.55 0.39 0.56 0.68 LF RMSE 31.79 29.57 33.03 36.76 34.22 28.90 34.26 27.80 32.41 40.90 41.39 32.30 CV 0.27 0.28 0.26 0.28 0.26 0.27 0.29 0.31 0.26 0.27 0.28 0.32 R2 0.59 0.55 0.51 0.53 0.48 0.67 0.71 0.63 0.81 0.70 0.62 0.85 TJ RMSE 30.65 28.28 33.59 33.58 34.08 25.31 28.62 26.11 29.69 24.56 34.28 26.76 CV 0.27 0.30 0.28 0.26 0.26 0.26 0.26 0.29 0.25 0.18 0.28 0.27 R2 0.58 0.57 0.54 0.51 0.47 0.61 0.71 0.52 0.65 0.47 0.96 0.71 BD RMSE 35.77 32.29 36.47 41.57 40.31 32.59 34.61 28.86 39.99 40.44 36.22 26.30 CV 0.29 0.31 0.28 0.30 0.28 0.30 0.28 0.27 0.30 0.27 0.25 0.26 R2 0.55 0.58 0.43 0.49 0.49 0.59 0.56 0.59 0.39 0.31 0.64 0.65 CZ RMSE 31.10 29.67 34.21 32.01 33.99 26.76 35.56 29.72 34.83 43.02 31.89 41.75 CV 0.25 0.27 0.25 0.23 0.24 0.24 0.28 0.26 0.25 0.30 0.22 0.43 R2 0.53 0.40 0.47 0.41 0.43 0.59 0.47 0.40 0.32 0.43 0.48 0.53 SJZ RMSE 36.26 34.06 36.82 39.96 40.73 33.05 42.13 36.58 47.60 46.38 47.18 30.80 CV 0.31 0.37 0.30 0.29 0.29 0.33 0.36 0.38 0.35 0.32 0.40 0.33 R2 0.46 0.41 0.37 0.38 0.26 0.49 0.50 0.36 0.29 0.61 0.39 0.52 TY RMSE 34.54 26.49 36.56 40.26 42.37 30.96 34.73 26.77 39.70 37.03 38.29 31.26 CV 0.39 0.39 0.41 0.38 0.37 0.42 0.41 0.38 0.42 0.39 0.37 0.46 R2 0.50 0.45 0.40 0.30 0.45 0.62 0.53 0.63 0.33 0.20 0.74 0.87 XT RMSE 36.49 35.48 35.13 45.06 43.03 27.06 37.65 31.59 40.98 43.99 45.58 24.43 CV 0.32 0.37 0.30 0.33 0.31 0.28 0.32 0.35 0.33 0.31 0.31 0.26 R2 0.45 0.43 0.35 0.18 0.34 0.54 0.50 0.54 0.40 0.17 0.57 0.53 HD RMSE 36.41 33.66 35.83 45.31 41.09 30.25 34.40 31.37 35.42 39.19 38.32 28.81 CV 0.31 0.35 0.28 0.34 0.31 0.31 0.31 0.33 0.28 0.29 0.33 0.32 R2 0.25 0.32 0.08 0.12 0.14 0.45 0.24 0.32 0.12 0.18 0.28 0.48 ZZ RMSE 47.22 39.60 50.19 55.78 57.67 35.32 48.88 40.72 49.63 64.64 55.83 35.87 CV 0.48 0.44 0.51 0.50 0.48 0.44 0.48 0.49 0.54 0.52 0.39 0.43 83 Figures

CD ZJK BJ QHD TS LF TJ BD CZ SJZ 2013 HS XT HD TY JN 0 ZZ CD ZJK BJ QHD TS LF 35 TJ BD CZ 2014 SJZ HS XT HD TY JN ZZ 75

CD ZJK BJ QHD TS LF TJ BD 115 CZ SJZ

2015 HS XT HD TY JN ZZ 250 CD ZJK BJ QHD TS LF TJ BD CZ 350 SJZ 2016 HS XT HD TY JN ZZ

CD ZJK BJ QHD TS LF TJ BD CZ 2017 SJZ HS XT HD TY JN ZZ 84 4/1 4/11 4/21 5/1 5/11 5/21 5/31 6/10 6/20 6/30 7/10 7/20 7/30 8/9 8/19 8/29 9/8 9/18 9/28 10/8 10/18 10/28

85 Fig. S1. Time series of daily averaged PM2.5 concentrations in 14 cities (north to south) during April to 86 October from 2013 to 2017. Six ranks are separated, representing different air-quality levels, including 87 excellent (green spots), good (yellow), lightly polluted (orange), moderately polluted (red), heavily polluted 88 days (purple) and severely polluted (maroon) with cut-off concentrations as 35, 75, 115, 150 and 250 μg m-3, 89 respectively.

27 1980-2017 2013-2017 24 2017

21

temperature(℃) 18

15 90 5/1 5/4 5/7 5/10 5/13 5/16 5/19 5/22 5/25 5/28 5/31

91 Fig. S2. Time series of the daily average temperature from 1980-2017, 2013-2017 and 2017. 92 93 Fig. S3. Wind field and boundary layer height (based on ERA-interim data) and occurrence days under 94 different weather conditions. The colour shading corresponds to the mean boundary layer height. The black 95 ‘*’ indicates the center of North China.

96 97 Fig. S4. Spatial distributions of average temperature for the 26 weather types. The first, second, and third 98 rows correspond to the N-E-S direction, S-W-N direction, and LP, respectively; the fourth row from left to 99 right is C and A, respectively. 100 101 Fig. S5. Same as Fig. S4 but for RH (%).

102 103 Fig. S6. Same as Fig. S4 but for rain (mm). 104 105 Fig. S7. Same as Fig. S4 but for TCC. 106 107 Fig. S8. Pressure field characteristics and occurrence days (n) of weather type N to CE in 2013-2017. 108 109 Fig. S9. Pressure field characteristics and occurrence days (n) of weather type CSE to ANW in 2013-2017. 110 111 Fig. S10. Comparison of the observed and predicted values in Beijing from April-October 2013-2017. 112 113 References

114 Demuzere, M., Trigo, R. M., Vila-Guerau de Arellano, J., and van Lipzig, N. P. M.: The impact of weather and 115 atmospheric circulation on O3 and PM10 levels at a rural mid-latitude site, Atmos. Chem. Phys., 9, 2695-2714, 116 https://doi.org/10.5194/acp-9-2695-2009, 2009. 117 Eder, B. K., Davis, J. M., and Bloomfield, P.: An Automated Classification Scheme Designed to Better Elucidate 118 the Dependence of Ozone on Meteorology, J.appl.meteor, 33, 1182-1199, 1994. 119 Han, H., Liu, J., Shu, L., Wang, T., and Yuan, H.: Local and synoptic meteorological influences on daily 120 variability of summertime surface ozone in eastern China, Atmospheric Chemistry and Physics Discussions, 1-51, 121 10.5194/acp-2019-494, 2019.

122 Huth, R.: An intercomparison of computer‐assisted circulation classification methods, International Journal of

123 Climatology: A Journal of the Royal Meteorological Society, 16, 893-922, 1996. 124 Huth, R., Beck, C., Philipp, A., Demuzere, M., Ustrnul, Z., Cahynová, M., Kyselý, J., and Tveito, O. E.: 125 Classifications of atmospheric circulation patterns, Annals of the New York Academy of Sciences, 1146, 105-152, 126 2008. 127 Jenkinson, A. F., Collison, F.P: An initial climatology of gales over the North Sea. , Synoptic Branch 128 Memorandum No. 62. Met Office, Exeter., 1977. 129 Jones, P. D., Hulme, M., and Briffa, K. R.: A comparison of Lamb circulation types with an objective 130 classification scheme, International Journal of Climatology, 13, 655-663, 1993. 131 Lamb, H. H.: British Isles weather types and a register of the daily sequence of circulation patterns, 1861–1971., 132 Geophysical Memoir., 116, p. 85., 1972. 133 Li, K., Jacob, D. J., Liao, H., Shen, L., Zhang, Q., and Bates, K. H.: Anthropogenic drivers of 2013–2017 trends in 134 summer surface ozone in China, Proceedings of the National Academy of Sciences, 116, 422-427, 135 10.1073/pnas.1812168116, 2019. 136 Liao, Z., Gao, M., Sun, J., and Fan, S.: The impact of synoptic circulation on air quality and pollution-related 137 human health in the Yangtze River Delta region, The Science of the total environment, 607-608, 838-846, 138 10.1016/j.scitotenv.2017.07.031, 2017. 139 Pope, R. J., Butt, E. W., Chipperfield, M. P., Doherty, R. M., Fenech, S., Schmidt, A., Arnold, S. R., and Savage, 140 N. H.: The impact of synoptic weather on UK surface ozone and implications for premature mortality, 141 Environmental Research Letters, 11, 124004, 10.1088/1748-9326/11/12/124004, 2016. 142 Santurtún, A., González-Hidalgo, J. C., Sanchez-Lorenzo, A., and Zarrabeitia, M. T.: Surface ozone concentration 143 trends and its relationship with weather types in Spain (2001–2010), Atmospheric Environment, 101, 10-22, 2015. 144 Trigo, R. M., and DaCamara, C. C.: Circulation weather types and their influence on the precipitation regime in 145 Portugal, International Journal of Climatology, 20, 1559-1581, 2000. 146 Yarnal, B.: Synoptic climatology in environmental analysis: a primer, Belhaven, 1993. 147